What Is a General-Purpose Technology? The Inventions That Rewire an Entire Economy

Most inventions do one thing well. A better plough turns soil; a vaccine prevents a disease; a stronger engine hauls a heavier load. But every so often a technology arrives that refuses to stay in its lane. It seeps into farming and finance, into warfare and art, and within a generation the things built on top of it look nothing like the things that came before. Economists have a deliberately unglamorous name for these rare inventions: general-purpose technologies.

The phrase belongs to a particular tradition in economic history, one that treats growth not as a smooth curve but as a sequence of upheavals organised around a few foundational tools. The claim is that long stretches of prosperity are not simply the sum of millions of unrelated improvements. They are downstream of a small number of technologies general enough to be useful almost everywhere — engines of growth, in the phrase the field itself adopted.

So what qualifies one? The classic account sets three tests. The first is pervasiveness: the technology spreads across most sectors instead of serving a single niche. The second is a long runway for improvement, so it keeps getting cheaper and more capable for decades and adopting it is never a one-time purchase. The third, and most important, is innovational complementarity — it makes other people’s inventions possible. Electricity did not merely light rooms; it enabled the assembly line, the home appliance, the elevator, and the dense cities that needed all three. A general-purpose technology is, in effect, a platform for other breakthroughs.

By those tests the canonical examples are familiar: the steam engine, electrification, the internal-combustion engine, the computer. But the list reaches far deeper into the past than the industrial era. Historians of technology count only a couple of dozen true general-purpose technologies across roughly ten thousand years, starting with things as elemental as agriculture, writing, and the wheel. MCM has argued that even rope belongs on that list — an unglamorous tool whose absence would have unmade most of what followed.

The most counter-intuitive feature of a general-purpose technology is that it disappoints first. Because the surrounding economy was built for the old way of doing things, the new tool delivers little until everything around it is redesigned. When factories first swapped steam engines for electric motors, they bolted the motors onto the same central drive shafts and saw almost no gain; the payoff came decades later, once plants were rebuilt around the idea that every machine could have its own power. The same lag produced the famous observation that the computer age was visible everywhere except in the productivity statistics. The dip-then-surge is now formalised as a “productivity J-curve”: measured output sags while firms make the slow, invisible investments in skills and reorganisation that the technology actually demands.

This is why the label is more than academic housekeeping. To call something a general-purpose technology is to make a claim about where decades of growth will come from — and a warning about how unevenly it tends to land. The gains accrue to whoever rebuilds fastest and reaches the new tools, which is why questions of access are economic and not merely moral; MCM has reported on how shutting half the workforce out of the next wave of automation quietly drains the growth it promises.

All of which frames the argument now consuming boardrooms and finance ministries: is artificial intelligence the next one? It looks the part. It is spreading across sectors, improving relentlessly, and already reshaping how ordinary people think, write, and work. Some economists go further and suspect AI is not only general-purpose but an “invention of a method of invention” — a tool that accelerates the discovery of everything else. It is also why governments and companies are pouring capital into the supercomputing infrastructure the technology runs on, betting on the platform before the economy built to exploit it fully exists.

If the pattern holds, the honest answer to whether AI is a general-purpose technology will arrive the way it always has: late, and only once we have rebuilt enough of the world around it to measure what changed. The engines of growth are rarely recognised while they are still warming up.

Tags: artificial intelligence, general-purpose technology